Add language detector, code analyzer, and issue detector

src/gdiffer/issue_detector.py

@@ -0,0 +1,101 @@
+"""Issue detector for common bugs, security vulnerabilities, and code smells."""
+
+import re
+from dataclasses import dataclass, field
+from typing import Optional
+
+
+@dataclass
+class Issue:
+    type: str
+    severity: str
+    title: str
+    description: str
+    line: Optional[int] = None
+    suggestion: str = ""
+    pattern: str = ""
+
+
+class IssueDetector:
+    SECURITY_PATTERNS = [
+        {'pattern': r'(?i)(sql\s*\(|execute\s*\(|exec\s*\()', 'type': 'sql_injection', 'severity': 'critical', 'title': 'Potential SQL Injection', 'description': 'String concatenation in SQL query', 'suggestion': 'Use parameterized queries'},
+        {'pattern': r'(?i)(innerHTML\s*=|outerHTML\s*=)', 'type': 'xss', 'severity': 'critical', 'title': 'Potential XSS Vulnerability', 'description': 'Directly setting HTML content', 'suggestion': 'Use textContent or sanitize HTML'},
+        {'pattern': r'(?i)(eval\s*\()', 'type': 'code_injection', 'severity': 'critical', 'title': 'Code Injection Risk', 'description': 'eval() detected', 'suggestion': 'Avoid eval()'},
+        {'pattern': r'(?i)(os\.system\s*\(|subprocess\.|shell=True)', 'type': 'command_injection', 'severity': 'critical', 'title': 'Command Injection Risk', 'description': 'Shell command execution', 'suggestion': 'Use subprocess with shell=False'},
+        {'pattern': r'(?i)(password\s*=|passwd\s*=|secret\s*=|token\s*=)', 'type': 'hardcoded_secret', 'severity': 'high', 'title': 'Hardcoded Secret', 'description': 'Potential hardcoded credential', 'suggestion': 'Use environment variables'},
+        {'pattern': r'(?i)(http://)', 'type': 'insecure_transport', 'severity': 'medium', 'title': 'Insecure HTTP', 'description': 'Using HTTP instead of HTTPS', 'suggestion': 'Use HTTPS'},
+        {'pattern': r'(?i)(random\.randint\s*\()', 'type': 'weak_crypto', 'severity': 'medium', 'title': 'Weak Random', 'description': 'Using random module', 'suggestion': 'Use secrets module'},
+    ]
+
+    BUG_PATTERNS = [
+        {'pattern': r'(?i)(if\s*\([^)]*==[^)]*\)\s*:)', 'type': 'assignment_in_condition', 'severity': 'high', 'title': 'Assignment in Condition', 'description': 'Possible typo = instead of ==', 'suggestion': 'Use == for comparison'},
+        {'pattern': r'(?i)(\bNone\b.*==)', 'type': 'none_comparison', 'severity': 'medium', 'title': 'Direct None Comparison', 'description': 'Using == None', 'suggestion': 'Use is None'},
+        {'pattern': r'\bexcept\s*:\s*$', 'type': 'bare_except', 'severity': 'medium', 'title': 'Bare Except Clause', 'description': 'Catching all exceptions', 'suggestion': 'Catch specific exceptions'},
+    ]
+
+    CODE_SMELL_PATTERNS = [
+        {'pattern': r'(?i)(\bTODO\b|\bFIXME\b)', 'type': 'code_tag', 'severity': 'low', 'title': 'Code Tag', 'description': 'TODO/FIXME comment', 'suggestion': 'Address or create ticket'},
+        {'pattern': r'(?i)(\bprint\s*\()', 'type': 'debug_statement', 'severity': 'low', 'title': 'Debug Statement', 'description': 'print() detected', 'suggestion': 'Remove debug statements'},
+        {'pattern': r'.{80,}', 'type': 'long_line', 'severity': 'low', 'title': 'Long Line', 'description': 'Line exceeds 80 characters', 'suggestion': 'Split long lines'},
+        {'pattern': r'\bpass\b', 'type': 'empty_block', 'severity': 'low', 'title': 'Empty Code Block', 'description': 'Empty pass statement', 'suggestion': 'Add explanatory comment'},
+    ]
+
+    def __init__(self):
+        self._compiled_patterns = []
+        self._compile_patterns()
+
+    def _compile_patterns(self):
+        self._compiled_patterns = []
+        for pattern_info in self.SECURITY_PATTERNS + self.BUG_PATTERNS + self.CODE_SMELL_PATTERNS:
+            try:
+                compiled = re.compile(pattern_info['pattern'])
+                self._compiled_patterns.append((compiled, pattern_info))
+            except re.error:
+                pass
+
+    def detect_issues(self, code: str, language: str = "text") -> list[Issue]:
+        issues = []
+        for line_num, line in enumerate(code.splitlines(), 1):
+            for compiled, pattern_info in self._compiled_patterns:
+                if compiled.search(line):
+                    issues.append(Issue(
+                        type=pattern_info['type'], severity=pattern_info['severity'],
+                        title=pattern_info['title'], description=pattern_info['description'],
+                        line=line_num, suggestion=pattern_info['suggestion'], pattern=pattern_info['pattern'],
+                    ))
+        return issues
+
+    def detect_diff_issues(self, old_code: str, new_code: str, language: str = "text") -> list[Issue]:
+        issues = []
+        for i, line in enumerate(new_code.splitlines(), 1):
+            if line.startswith('+') and not line.startswith('+++'):
+                clean_line = line[1:]
+                for compiled, pattern_info in self._compiled_patterns:
+                    if compiled.search(clean_line):
+                        issues.append(Issue(
+                            type=pattern_info['type'], severity=pattern_info['severity'],
+                            title=pattern_info['title'], description=pattern_info['description'],
+                            line=i, suggestion=pattern_info['suggestion'], pattern=pattern_info['pattern'],
+                        ))
+        return issues
+
+    def suggest_improvements(self, code: str, language: str = "text") -> list[str]:
+        suggestions = []
+        issues = self.detect_issues(code, language)
+        severity_order = {'critical': 0, 'high': 1, 'medium': 2, 'low': 3}
+        seen_types = set()
+        for issue in sorted(issues, key=lambda x: (severity_order.get(x.severity, 4), x.title)):
+            if issue.type not in seen_types and issue.suggestion:
+                suggestions.append(f"{issue.title}: {issue.suggestion}")
+                seen_types.add(issue.type)
+        return suggestions
+
+
+def detect_issues(code: str, language: str = "text") -> list[Issue]:
+    detector = IssueDetector()
+    return detector.detect_issues(code, language)
+
+
+def suggest_improvements(code: str, language: str = "text") -> list[str]:
+    detector = IssueDetector()
+    return detector.suggest_improvements(code, language)